2026-10-27 20:04:12
Welcome to the Premium edition of How They Make Money.
Over 300,000 subscribers turn to us for business and investment insights.
In case you missed it:
The company behind Claude is growing faster than we’ve ever seen in software.
Founded only five years ago, it’s already challenging the world’s largest technology companies for the future of AI.
Anthropic finally filed its S-1, revealing a business that generated $16.3 billion in revenue in the first half of 2026, up 14x year-over-year. The Financial Times reported that revenue reached $16.8 billion in Q3 alone.
The investment thesis comes down to three things:
Claude is winning enterprise customers.
AI agents are expanding what software can do.
Compute costs could make or break the economics.
Anthropic may be building one of the most valuable software platforms of the next decade.
But staying at the frontier of AI comes at an unprecedented cost.
I dug through 300+ pages of its S-1 to unpack the business, the financials, and the risks, supported by our signature visuals. By the end, you’ll understand how Anthropic makes money, what its extraordinary growth actually costs, and how to interpret its valuation when the IPO arrives.
Today at a glance:
Overview
Business model
Financial highlights
Risks & challenges
Management
Use of proceeds
Future outlook
Personal take
Anthropic started as an AI safety research lab. Five years later, it’s one of the leading challengers to OpenAI and Google, with Claude increasingly embedded in the operations of some of the world’s largest companies.
Founded in 2021 by seven co-founders, including former OpenAI researchers, Anthropic was built around the belief that increasingly powerful AI systems must also become safer and more reliable. That mission remains central to its identity, but it has also become a commercial selling point: Enterprises need AI they can trust with sensitive data and business-critical work.
Headquarters: San Francisco, California
Founded: 2021 by Dario and Daniela Amodei and five other co-founders
Ticker: ANT (proposed Nasdaq listing)
Mission: Responsible development and maintenance of advanced AI for the long-term benefit of humanity
Structure: Public benefit corporation (PBC)
The easiest way to understand Anthropic’s ambition is to look beyond the chatbot.
Most people discovered generative AI through a chat window: Ask a question, get an answer. Anthropic believes that’s just the beginning.
Its S-1 introduces the concept of “Virtual Collaborators” — AI systems capable of planning, reasoning, navigating software, and completing complex tasks autonomously. Instead of merely suggesting what a human should do next, Claude can increasingly execute the work itself.
Claude Code demonstrated this shift for software developers, moving from code suggestions to writing, testing, and debugging entire projects. Claude Cowork extends the same approach to other knowledge workers, handling files, applications, and multi-step workflows.
The ambition is to transform Claude from a tool people consult into a colleague they delegate work to.
Anthropic had approximately 20 customers spending at least $100K annually at the end of 2023. That figure reached roughly 200 in 2024, 1,500 in 2025, and more than 11,900 by September 2026.
And these aren’t just small experiments:
More than 1,900 customers generated at least $1 million in trailing annual revenue.
Approximately 250 customers generated at least $10 million.
Every Fortune 10 company and more than half of the Fortune 500 were customers spending at least $100,000 annually.
These are actual revenue thresholds. Anthropic has moved remarkably quickly from winning individual developers to becoming a meaningful vendor inside the world’s largest enterprises.
The business is also becoming less concentrated. Anthropic’s top five customers accounted for 21% of Q3 revenue, down from 36% in Q1, even as spending from each of those major customers hit new highs.
But one S-1 metric stands out: 900% net revenue retention.
Understanding what drives that expansion, whether it can last, and how much of it Anthropic can turn into profits is central to the investment case.
2026-10-24 22:00:57
Welcome to the Saturday PRO edition of How They Make Money.
Over 300,000 subscribers turn to us for business and investment insights.
In case you missed it:
📊 Monthly reports: 200+ companies visualized.
📩 Tuesday articles: Exclusive deep dives and insights.
📚 Access to our archive: Hundreds of business breakdowns.
📩 Saturday PRO reports: Timely insights on the latest earnings.
Today at a glance:
🕶️ Meta: AI Bet Takes Shape
🚖 Tesla: $5 Billion CapEx Reversal
🧴 P&G: Volume Returns
🧠 Lam Research: AI's Storage Problem
☁️ SAP: Backlog Redemption
⚙️ Texas Instruments: Industrial Breaks Out
🌐 IBM: Mainframe Comeback
💳 Amex: Premium Trade-Off
🧬 Abbott: A Leaner Future
📞 AT&T: Cost of Convergence
🦾 Intuitive Surgical: Cost of Expansion
🛰️ Lockheed Martin: $292 Billion Backlog
🎧 Spotify: Agentic Advantage
💼 Moody’s: AI Financing Boom
🚗 GM: Cash Over Growth
🛩️ United: Pricing Power Meets Its Limits
🛩️ Southwest: Fuel Squeeze
🦅 American Airlines: Fuel Trap
🏡 Appfolio: AI Goes to Work
Meta’s Q3 revenue rose 26% Y/Y to a record $64.6 billion ($1.2 billion beat), while adjusted EPS reached $7.40 ($0.60 beat). GAAP EPS of $3.83 included a $10 billion legal charge tied to its settlement of youth-safety lawsuits.
Shares fell 7% as investors looked past strong advertising growth and encouraging early traction for Muse, and focused instead on mounting AI spending and deteriorating free cash flow.
Ad revenue grew 26% Y/Y to $63.1 billion, with impressions up 10% and average price per ad rising 14%. That’s a shift from Q2, when impression growth (14%) outpaced pricing growth (12%). Meta is increasingly monetizing its existing audience more effectively rather than simply showing more ads. Family daily active people reached 3.64 billion (+3% Y/Y), while AI-powered recommendations and targeting continued to support engagement and conversion rates.
But the bigger story is Muse becoming a tangible business opportunity just six weeks after launch:
More than 10 million users have tried Muse in the US and Canada, with approximately 25% returning on a given day. That’s promising early engagement, although retention and monetization remain unproven at scale.
Shopping is the leading use case, with more than 25% of users engaging in commerce activities. Meta sees opportunities in subscriptions, transaction fees, and eventually advertising, extending its reach from influencing purchases to facilitating them.
Muse Wonder, Meta’s first family of frontier models, is already powering new capabilities. Management expects to improve Muse’s compute efficiency by more than 10x over the next year, an important step toward making the agent economically viable.
Zuck believes Muse could eventually serve billions of people and generate more revenue per user than Meta’s existing social platforms. The ambition is enormous, but so are the obstacles. Amazon and other websites have blocked Muse from shopping on their platforms, illustrating why Meta is developing a Personal Agent Protocol to enable structured interactions between agents and businesses. Winning the interface doesn’t automatically guarantee access to the transaction.
Meanwhile, the AI bill keeps growing:
Free cash flow turned negative for the first time, at -$2.1 billion, compared with $784 million in Q2 and $10.6 billion a year ago. Capital expenditures reached $34.9 billion, up from $31.1 billion in Q2.
Meta raised the floor of FY26 CapEx guidance again, to $135–145 billion (from $130–145 billion). Even excluding the legal charge, operating margin declined to roughly 35% from 40% a year ago as AI infrastructure depreciation and R&D spending increased.
Reality Labs lost another $4.1 billion. Management now expects annual losses of $18–19 billion for the segment in FY26 and approximately $15 billion in 2027 as resources shift toward AI.
Meta also announced a new class of non-voting shares, effectively creating a 2-for-1 stock split in January 2027. The Class C shares could fund acquisitions, employee compensation, or future capital raises without diluting Zuck’s voting control. It’s a notable move for a company that historically funded its ambitions through advertising cash flow.
More legal risks loom. Beyond the $10 billion Q3 charge, a New Mexico judge ordered Meta to pay $25 billion over alleged consumer deception related to privacy and content moderation. Meta plans to appeal, and the judgment is separate from the previously announced multistate settlement.
Meta guided Q4 revenue to $73–76 billion (midpoint above consensus) and raised FY26 expense guidance to $175–179 billion, incorporating the legal charge. Management still expects FY26 operating income, excluding that charge, to exceed last year’s level.
Takeaway: Meta’s ad business remains remarkably strong, and Muse offers a more tangible path toward a new revenue stream. But AI spending now exceeds operating cash flow. Can Muse become a meaningful business before infrastructure costs put further pressure on margins? Early adoption is encouraging, but monetization remains unproven.
Tesla’s Q3 offered a mixed picture. Automotive margins recovered, and FSD adoption accelerated. But vehicle sales remain stagnant, while heavy investment in AI and robotics continues to weigh on profitability.
The biggest surprise? Tesla returned to positive free cash flow after burning cash last quarter, helped by a significant slowdown in capital expenditures.
Revenue was flat Y/Y at $28.1 billion ($0.2 billion miss).
Gross margin reached 20% (+2pp Y/Y and +3pp Q/Q).
Operating margin fell to 3% (-3pp Y/Y).
Non-GAAP EPS was $0.48 ($0.03 beat).
Operating cash flow reached $6.3 billion, while free cash flow fell 62% Y/Y to $1.5 billion as CapEx declined Q/Q from $5.8 billion to $4.7 billion.
Tesla lowered its FY26 CapEx outlook from more than $25 billion to ~$20 billion. With $13 billion spent through September, that implies roughly $7 billion in Q4, still a meaningful acceleration from Q3.
Management expects spending to grow beyond 2026 as Tesla expands Robotaxi, Optimus, AI compute, solar, and semiconductor manufacturing. Tesla ended Q3 with $45.5 billion in cash and investments, plus about $30 billion in borrowing capacity it does not expect to draw on this year.
Now, here are the main insights from the call:
2026-10-23 20:03:42
Welcome to the Free edition of How They Make Money.
Over 300,000 subscribers turn to us for business and investment insights.
In case you missed it:
Netflix has proved that streaming can generate enormous profits. But can it sustain growth when viewers have more entertainment choices than ever?
Management admits recent execution has been hit or miss, and the company is replacing longtime content chief Bela Bajaria with former Disney executive Peter Rice, signaling a renewed push for watercooler hits.
The stock price has dropped more than 45% over the past year, and the company recently announced its largest round of layoffs since 2022, eliminating roughly 1,000 positions.
Netflix is on track to surpass $50 billion in annual revenue, generates billions in free cash flow, and remains the world's largest paid streaming platform. Yet its share of US television viewing has plateaued, while YouTube and other competitors continue to gain share on TV.
Alongside its core programming, Netflix is experimenting with new ways to capture consumer attention. Live sports, video podcasts, creator content, and partnerships with traditional broadcasters offer new ways to attract and retain subscribers.
Today at a glance:
🍿 Netflix Q3 FY26
📈 YouTube pulls away
📺 Not all viewing hours are equal
💬 Salesforce buys Braze for $5.7B
Revenue +12% Y/Y to $12.9 billion (in line).
Operating margin 35% (+6pp Y/Y).
EPS +49% Y/Y to $0.87 ($0.05 beat).
Cash and short-term investments: $8.2 billion.
Debt: $15.3 billion.
Revenue +13% to $51.1 billion (slightly below the prior midpoint).
Operating margin 31.5% (+2pp Y/Y, unchanged).
Free cash flow ~$12.0 billion (down from $12.5 billion).
📉 Growth keeps slowing: Revenue growth decelerated to 12%, its slowest pace in nearly three years, with Q4 guidance implying roughly the same rate. Operating margin surged 6 points to 35%, although the beat reflected favorable expense timing. Management reiterated its 31.5% full-year margin target despite restructuring costs in Q4. Subscription revenue is outperforming expectations, supported by price increases, membership growth, and improving churn. But margin expansion alone may no longer satisfy investors.
📢 Ads miss the mark: Netflix cut its 2026 advertising revenue forecast from ~$3 billion to ~$2.7 billion, blaming a slower-than-expected rollout of ad products. That represents a 10% cut, even though ad revenue is still expected to grow over 80% Y/Y. Netflix continues to expand programmatic access and measurement capabilities, with plans to introduce its ad tier in 15 additional countries in 2027. The opportunity remains substantial, but future growth will depend on technical execution and sustained viewer engagement.
🌍 Becoming an entertainment aggregator: Netflix is expanding beyond its own productions by integrating third-party programming directly into its service. Its partnership with French broadcaster TF1 is already generating incremental viewing, while a new Disney licensing agreement brings titles such as Ice Age to the platform. Management is open to additional integrations, potentially including subscriptions to other streaming services. The irony is that Disney is using Netflix to build awareness for its own upcoming theatrical releases. Even Hollywood's largest studios recognize Netflix's distribution value. Netflix may not always have the best original programming, but its distribution advantage is difficult to replicate. That could help the company expand its entertainment offering without continually increasing content spending.
💰 Record buybacks continue: Netflix repurchased another $4.8 billion of stock, its largest quarterly buyback ever, leaving $22.3 billion in remaining authorization. Free cash flow rose to $2.9 billion from $2.7 billion a year ago. Management reduced its full-year FCF forecast from $12.5 billion to about $12 billion, mainly because a real estate transaction is expected to close in Q4. Netflix continues to generate substantial cash and return it to shareholders, even as revenue growth moderates.
Streaming keeps taking share from traditional television, but Netflix isn’t keeping pace. According to Nielsen, streaming reached a record 49.0% of US TV viewing in July, up from 47.3% a year ago.
Netflix captured 7.8% of US TV time, down from 8.8% a year earlier. YouTube, meanwhile, reached a record 14.2%, widening its lead over Netflix from 4.6 to 6.4 percentage points in just 12 months. Americans now spend nearly twice as much time watching YouTube as Netflix on their TVs. And that doesn't even include YouTube TV.
Netflix had an unusually strong summer last year, when Squid Game and other hits dominated the charts. Monthly viewing fluctuates with release schedules, and Nielsen only measures US television screens (excluding mobile and laptop viewing). Still, the trend highlights how difficult it is to sustain engagement against free services and platforms built around seemingly endless creator content.
Netflix has acknowledged the challenge. Co-CEO Ted Sarandos said that engagement wasn’t growing as quickly as he wanted. New content chief Peter Rice previously oversaw Fox Searchlight, which delivered hits like Slumdog Millionaire, before taking senior leadership roles at Disney. His appointment suggests Netflix wants to improve its ability to produce breakout US hits that can travel internationally.
Yet Netflix’s own engagement data paints a more nuanced picture. Worldwide viewing hours rose 2% in H1 2026, up from 1.5% growth in 2025. Netflix hasn’t disclosed Q3 viewing hours, but management reported improving churn globally and in the US. Engagement continued to grow even as Netflix’s share of US television viewing declined. In short, subscriber retention remains healthy, but subscriber loyalty masks a gradual erosion of its share of screen time.
Netflix argues that viewing hours alone don’t capture the value of its programming. Its live events illustrate why. Live programming represents just over 5% of Netflix’s content spending but only about 1% of viewing hours. Yet live events drove six of its 10 biggest sign-up days over the past five years.
That makes sports potentially more valuable than their modest engagement contribution suggests. A major NFL game or boxing match can attract new subscribers, create premium advertising inventory, and generate cultural relevance. The economics depend less on how long viewers watch than on whether they remain subscribers afterward.
Netflix is also exploring lower-cost ways to become a daily entertainment destination. Its partnership with French broadcaster TF1 achieved the broadcaster’s 18-month audience targets in less than three weeks. TF1+ daily audiences increased 16% during launch week, while Netflix reported growing viewing hours for TF1 programming. The early results are encouraging, although they don’t yet establish how much incremental value Netflix captures.
Video podcasts and creator-led programming offer another way to expand viewing occasions without the production costs of premium scripted entertainment. Netflix is even experimenting with theatrical releases. Greta Gerwig’s upcoming Narnia adaptation will receive a wide theatrical release before arriving on the service, testing whether the big screen can help build a streaming franchise.
Still, management isn’t abandoning its core business. Roughly 95% of spending remains focused on its core streaming offering, including series, films, and live events. These newer initiatives complement the core business rather than represent a fundamental shift in strategy.
Netflix’s challenge is to capture more valuable viewing occasions, not necessarily the most viewing hours. A blockbuster can drive sign-ups, a licensed show can improve retention, and live sports can attract advertisers. Each serves a different purpose.
Management points to improving churn and successful price increases as evidence of healthy engagement. But churn could be a lagging indicator. Subscribers could gradually spend less time watching Netflix long before they consider canceling.
Netflix has proven it can raise prices and expand margins. Now it needs to show that it can grow without relying primarily on charging existing subscribers more.
Salesforce agreed to acquire customer engagement platform Braze for $5.7 billion in enterprise value, offering $47.50 per share in cash, a 58% premium to its previous closing price. The acquisition raises an obvious question: Why does Salesforce need Braze when it already owns Marketing Cloud?
Braze specializes in real-time customer engagement. Rather than relying on scheduled marketing campaigns, its software reacts to customer behavior across mobile apps, websites, email, and other channels. Think of an abandoned shopping cart triggering a personalized offer, or a streaming service reaching out to a subscriber showing signs of leaving. Customers include Canva, Instacart, and Venmo.
Braze’s Q2 FY27 results illustrate both the appeal and the challenges of the business:
Revenue grew 26% Y/Y to $227 million, with subscriptions accounting for 92%.
Dollar-based net retention reached 110%, indicating continued expansion among existing customers.
Sales and marketing consumed 41% of revenue, contributing to an $18 million operating loss despite a healthy 67% gross margin.
At roughly 6.2x annualized revenue, the acquisition isn’t outrageously expensive for a growing software business. Braze also generated $22 million in quarterly free cash flow, suggesting its underlying economics are improving despite its GAAP losses.
The strategic appeal is clear. Salesforce has customer data, enterprise relationships, and its Agentforce AI platform. Braze adds a specialized engagement engine that can translate those insights into personalized actions across consumer-facing channels. AI agents become more valuable when they can act on customer information. Salesforce could also accelerate Braze’s adoption through its massive enterprise distribution network.
But there are risks. Braze overlaps with Salesforce’s existing Marketing Cloud capabilities, creating potential integration challenges and product confusion. Its appeal as an independent platform could also suffer if customers fear being pushed into Salesforce’s broader ecosystem. And while cross-selling could improve margins, Salesforce must demonstrate that another acquisition actually accelerates organic growth.
The bigger picture is that Salesforce is trying to own more of the customer relationship, from data and AI-driven decisions to the interactions that follow. Braze could help close that loop. But can the integration deliver enough synergies and additional growth to justify the purchase price?
Disclosure: I own CRM and NFLX in App Economy Portfolio. I share my ratings (BUY, SELL, or HOLD) with App Economy Portfolio members.
Author's Note (Bertrand here 👋🏼): The views and opinions expressed in this newsletter are solely my own and should not be considered financial advice or any other organization's views.
2026-10-21 07:49:37
Welcome to the Premium edition of How They Make Money.
Over 300,000 subscribers turn to us for business and investment insights.
In case you missed it:
SpaceX just delivered its first profitable quarter as a public company.
Revenue surged 161% to $14.6 billion, and net income reached $4.4 billion. The rebranded Super Intelligence (SI) segment now generates more revenue than Starlink, transforming the rocket manufacturer into an AI powerhouse. But that growth came with nearly $20 billion in quarterly CapEx.
An even bigger move may be coming soon.
During the earnings call, Musk confirmed that SpaceX and Tesla have formed special committees to explore a merger. A combination could create a company worth more than $3 trillion, bringing Musk's AI, robotics, and space ambitions under one roof.
But would a merger unlock greater value for shareholders, or simply combine two enormous investment programs with very different economics?
Today at a glance:
SpaceX Q3 FY26.
Starlink opens new fronts.
Key insights from the call.
SpaceX + Tesla: One Musk empire?
SpaceX beat Wall Street expectations in Q3, with revenue $1.6 billion ahead and GAAP EPS of $0.35 ($0.19 beat). Adjusted EBITDA quadrupled Y/Y to $8.8 billion, driven by growth across Connectivity and SI.
The three businesses are moving at very different speeds:
🚀 Space revenue grew 6% Y/Y to $1.8 billion (+84% Q/Q) as SpaceX completed 12 customer launches. The segment roughly broke even despite continued investment in Starship, which generated revenue for the first time. Its successful deployment of next-generation Starlink satellites marks an important step toward commercial operations.
🛰️ Connectivity revenue jumped 69% Y/Y to $5.2 billion (+22% Q/Q), supported by consumer growth (+57%) and enterprise and government demand (+91%). Starlink subscribers reached 13.6 million (+89% Y/Y), but average revenue per user fell 17% (+5% Q/Q) to $69 as the service expanded internationally. Operating margin improved 7 percentage points to 45%.
🤖 Super Intelligence revenue soared 830% Y/Y to $7.6 billion (+195% Q/Q), becoming SpaceX's largest segment with 52% of total revenue. Growth reflected new cloud computing agreements, increased Grok adoption, and the acquisition of AI coding platform Cursor, which contributed $608 million in revenue from its August 14 acquisition through September 30. The segment swung from a $1.7 billion operating loss a year ago to $431 million in operating profit, although its 6% margin remains far below Starlink's 45%.
Operating cash flow reached $9.6 billion in Q3 (+156% Y/Y).
CapEx reached $19.6 billion, up from $18.4 billion in Q2. And management expects a much bigger jump in Q4.
Free cash flow was negative ~$10.0 billion, compared with negative ~$16.0 billion in Q2.
Super Intelligence accounted for $16.4 billion, or 84% of total CapEx in Q3, compared with $1.6 billion for Space and $1.5 billion for Connectivity. AI compute capacity increased from 1.4 GW to 1.7 GW during the quarter.
The operating cash flow improvement is encouraging, but SpaceX still invests roughly twice as much as it generates from operations.
Cash, cash equivalents, and marketable securities: $89.9 billion, down from approximately $100 billion at the end of Q2.
Debt and finance leases: $38.8 billion, including $36.1 billion of long-term obligations.
Net cash: approximately $51.0 billion, before other financing obligations or restricted-cash adjustments.
2026-10-17 22:01:07
Welcome to the Saturday PRO edition of How They Make Money.
Over 300,000 subscribers turn to us for business and investment insights.
In case you missed it:
📊 Monthly reports: 200+ companies visualized.
📩 Tuesday articles: Exclusive deep dives and insights.
📚 Access to our archive: Hundreds of business breakdowns.
📩 Saturday PRO reports: Timely insights on the latest earnings.
Today at a glance:
🔬 ASML: Demand Reaches 2030
💊 J&J: MedTech Needs Work
💼 UnitedHealth: Turnaround Broadens
🇺🇸 Bank of America: Fine Masks NII Growth
👔 Morgan Stanley: IPOs Feed Wealth
📈 BlackRock: Fee Growth Hits 11%
🏦 Schwab: Lending Boom
🍕 Domino’s: Orders Up, Check Down
ASML’s Q3 revenue surged 55% Y/Y to a record €11.7 billion (€70 million miss), while GAAP EPS reached €12.93 (€2.14 beat). The EPS beat was partly flattered by a €0.5 billion valuation gain on ASML’s investment in Mistral AI, but the underlying quarter was still impressive. Gross margin reached a record 57%, well above guidance, as high-margin upgrades pushed Installed Base Management sales to €3.3 billion.
The biggest insight was how far out the AI capacity cycle has moved. Customers are increasingly placing orders for 2028 and asking ASML to accept orders for 2029 and even 2030. Management now says its currently planned supply will not fully meet demand in either 2027 or 2028. That is remarkable because ASML already planned to increase Low-NA EUV and immersion DUV capacity by roughly 30% in 2027, followed by another ~30% increase in 2028. It is now exploring an even larger ramp in 2028.
The demand is increasingly broad:
Memory system sales are expected to grow ~75% in FY26, driven largely by DRAM capacity and greater EUV adoption.
Advanced logic is growing more than 25%, even though total logic growth is closer to 10%.
EUV sales should grow more than 45% this year, while Installed Base Management is now expected to grow roughly 50%.
ASML shipped 113 new lithography systems in Q3, up from 86 in Q2.
ASML has now secured price increases on some of its most critical high-volume tools. Management expects the benefit to begin showing up in 2027 and become fully visible in 2028. Management stressed this isn’t opportunistic annual repricing, but it does confirm that ASML can capture more of the value its bottleneck technology creates.
High-NA is moving closer to actual production. Intel has now processed more than one million wafers using High-NA, Samsung plans high-volume adoption in 2028, and TSMC now targets 2030. ASML expects High-NA usage to expand node by node as customers validate the economics.
China remains the geopolitical wildcard. It represented 23% of Q3 system sales, while ASML still expects roughly 20% of full-year revenue from China. Management argues underlying Chinese demand remains strong, but further restrictions on immersion DUV could eventually accelerate development of domestic alternatives.
ASML raised FY26 revenue guidance again to €44–€45 billion (from €43–€45 billion) and now expects a gross margin of roughly 56%.
Takeaway: ASML has moved from ramping capacity to figuring out how much further it can stretch the supply chain. Customers are already signaling demand into 2030, while pricing increases and High-NA add further upside. Now it’s about ASML’s ability to build enough machines.
2026-10-16 20:02:01
Welcome to the Free edition of How They Make Money.
Over 300,000 subscribers turn to us for business and investment insights.
In case you missed it:
Financing is the latest obstacle in the AI infrastructure boom.
CoreWeave shares plunged more than 15% on Wednesday, while yields on some of its bonds have climbed above 16%, a level associated with distressed debt. The NVIDIA-backed cloud provider has taken on tens of billions of dollars in debt to build the data centers powering the AI boom.
The pressure extends beyond CoreWeave. Oracle and other AI infrastructure companies have also faced selling pressure as investors reassess their borrowing costs and capital requirements. Higher interest rates make an already expensive buildout even more costly.
For highly leveraged AI infrastructure providers, the concern is shifting from demand for computing power to the cost of supplying it. Companies are committing enormous amounts of capital today against revenue they expect to collect over many years.
NVIDIA CEO Jensen Huang pushed back on Thursday, insisting that CoreWeave’s financial position is improving. But NVIDIA owns roughly 10% of CoreWeave and is also one of its biggest suppliers. Huang has every incentive to defend its expansion, even as bond investors demand higher returns to finance it.
Today at a glance:
⚡ TSMC is not slowing down
🤖 The model race heats up
🏠 Apple moves into the living room
Its US-listed shares have returned more than 100x since 1997.
I put together a free 35-page report on what to look for in quality stocks before they become obvious.
Taiwan Semiconductor (TSMC) is investing at a scale few companies can match. Its CapEx budget has climbed from $41 billion in 2025 to $60–$64 billion in 2026, with cumulative spending over 2026–2028 expected to substantially exceed the previous three years.
CEO C.C. Wei has little interest in leaving opportunities to competitors:
“We have no intention to leave food on the table for others, and we want to capture every piece of business possible.”
Another record quarter backs his conviction. Q3 Revenue grew 41% Y/Y to $46.7 billion, and operating margin reached nearly 59%. TSMC is becoming more profitable even as it undertakes the largest manufacturing expansion in its history.
Income statement:
Revenue +41% Y/Y to $46.7 billion ($0.9 billion above the high end of guidance).
Gross margin 66% (-1pp Q/Q).
Operating margin 59% (+8pp Y/Y).
EPS per ADR $4.60 (+69% Y/Y in local currency).
Cash flow:
Free cash flow margin: 32% (+18pp Y/Y).
Balance sheet:
Cash and marketable securities: ~$126 billion.
Interest-bearing debt: ~$32 billion.
Guidance:
Management raised its FY26 revenue growth outlook to approximately 44% in USD, up from slightly above 40%. Q4 guidance of $52.3–$54.3 billion implies 58% Y/Y growth at the midpoint, a sharp acceleration.
🤖 AI demand keeps surprising management. TSMC previously forecast mid-to-high-50% annual growth in AI accelerator revenue through 2029. C.C. Wei now suggests even that ambitious forecast could prove conservative:
“The demand is extremely robust. We did not see any sign of slowdown. Whether it is a high 50 CAGR, it does not matter. It probably will be higher.”
The forecast excludes conventional CPUs and networking chips, which are also benefiting from the rise of AI agents. TSMC’s opportunity extends well beyond GPUs, as its changing revenue mix shows. HPC (High Performance Computing) now accounts for 68% of revenue, up from 42% just three years ago.
🏗️ TSMC wants every available order. Yet the company isn’t simply taking its customers’ forecasts at face value. It checks their financial strength, visits data center construction sites, and verifies electricity availability before committing additional capacity. These checks help explain why management remains confident enough to invest tens of billions in factories that take years to build. They also connect directly to the financing concerns facing some AI infrastructure providers.
⚙️ Manufacturing leadership comes at a cost. The new 2nm process already accounts for 8% of wafer revenue, with capacity expected to more than double by 2027. Alongside advanced packaging and 3nm expansion, this reinforces TSMC’s position as the preferred manufacturer for the most demanding chips. However, the initial 2nm ramp contributed to the sequential decline in gross margin. Overseas expansion will add further pressure, with international factories expected to dilute consolidated gross margin by 3–4 percentage points over the next few years.
💰 How profitable can TSMC become? Despite rising investment and manufacturing costs, operating margin has climbed to nearly 59%. When asked about the longer-term potential, Wei replied:
“There is no ceiling for TSMC’s profitability. But a dramatic increase will not happen overnight either.”
TSMC’s technological lead and scarce capacity give it considerable pricing power. But management is also warning that rising component costs and macroeconomic uncertainty are hurting price-sensitive consumer markets. The company enters 2027 with exceptional AI demand, but not every part of the semiconductor industry is seeing the same conditions.
Google and Meta made major moves in the AI model race this week, challenging OpenAI and Anthropic's perceived lead.
Google launched Gemini 4 Argon, its latest frontier model, claiming leadership across several benchmarks involving coding, finance, law, and cybersecurity. Its performance on professional tasks (Vals Index) is particularly noteworthy as AI agents move beyond answering questions to performing actual work.
Argon is available through Google’s API, Gemini app, Search, and coding tools. Google can distribute increasingly capable AI to billions of existing users, an advantage over independent AI labs that must build their own distribution.
Meanwhile, Meta introduced Muse Wonder and Muse Wonder Full, its most capable models to date. After struggling to keep pace at the frontier, Meta claims competitive performance with OpenAI and Anthropic. The models emphasize memory and the ability to carry out complex, multi-step tasks. Meta plans to bring these capabilities to its Muse personal AI agent.
More competition could pressure model pricing, while more capable agents could increase computing demand. Unlike chatbots that answer individual questions, agents can work through lengthy tasks involving dozens or hundreds of steps, consuming far more tokens along the way.
Better models and lower pricing don’t necessarily mean less infrastructure spending. They could make AI useful enough to justify even more investment.
Apple unveiled HomeView, a $299 smart-home display arriving November 13. It’s built around an upgraded Siri, alongside new HomePod and Apple TV devices. The seven-inch display recognizes household members and draws on personal information from their iPhones to answer questions and handle everyday tasks.
Apple also introduced cameras, thermostats, and smart locks developed with LG and Schneider Electric. These accessories work directly through Apple Home, without separate apps or accounts. The new Apple TV adds far-field microphones for hands-free Siri commands.
Apple is betting that its hardware ecosystem and access to personal data can make Siri the default interface for the connected home. But there’s a catch: HomeView still relies on a nearby, recently unlocked iPhone for personalized features, and many of its use cases already work on the phone itself.
The real test is whether Apple’s integration can compensate for Siri’s limitations. Amazon and Google already have established smart-home ecosystems, while AI agents are becoming increasingly capable of handling tasks without traditional apps.
That's it for today.
Happy investing!
Disclosure: I own AAPL, AMZN, GOOG, META, and TSM in App Economy Portfolio. I share my ratings (BUY, SELL, or HOLD) with App Economy Portfolio members.
Author's Note (Bertrand here 👋🏼): The views and opinions expressed in this newsletter are solely my own and should not be considered financial advice or any other organization's views.